Tangent Spaces

Chapter 6
Manifolds, Tangent Spaces, Cotangent

Spaces, Vector Fields, Flow, Integral
Curves
6.1 Manifolds
In a previous Chapter we defined the notion of a manifold

embedded in some ambient space, RN .
In order to maximize the range of applications of the the-

ory of manifolds it is necessary to generalize the concept
of a manifold to spaces that are not a priori embedded in
some RN .
The basic idea is still that, whatever a manifold is, it is

a topological space that can be covered by a collection of
open subsets, Uα, where each Uα is isomorphic to some
“standard model”, e.g., some open subset of Euclidean
space, Rn.
305
306 CHAPTER 6. MANIFOLDS, TANGENT SPACES, COTANGENT SPACES
Of course, manifolds would be very dull without functions

defined on them and between them.
This is a general fact learned from experience: Geometry

arises not just from spaces but from spaces and interesting
classes of functions between them.
In particular, we still would like to “do calculus” on our

manifold and have good notions of curves, tangent vec-
tors, differential forms, etc.
The small drawback with the more general approach is

that the definition of a tangent vector is more abstract.
We can still define the notion of a curve on a manifold,

but such a curve does not live in any given Rn, so it it
not possible to define tangent vectors in a simple-minded
way using derivatives.
6.1. MANIFOLDS 307
Instead, we have to resort to the notion of chart. This is

not such a strange idea.
For example, a geography atlas gives a set of maps of var-

ious portions of the earh and this provides a very good
description of what the earth is, without actually imag-
ining the earth embedded in 3-space.
Given Rn, recall that the projection functions,

pri: Rn → R, are defined by
pri(x1, . . . , xn) = xi, 1 ≤ i ≤ n.
For technical reasons, from now on, all topological spaces

under consideration will be assumed to be Hausdorff and
second-countable (which means that the topology has a
countable basis).
Definition 6.1.1 Given a topological space, M , a chart

(or local coordinate map) is a pair, (U, ϕ), where U is
an open subset of M and ϕ: U → Ω is a homeomorphism
onto an open subset, Ω = ϕ(U ), of Rnϕ (for some
nϕ ≥ 1).
For any p ∈ M , a chart, (U, ϕ), is a chart at p iff p ∈ U .

If (U, ϕ) is a chart, then the functions xi = pri ◦ ϕ are
called local coordinates and for every p ∈ U , the tuple
(x1(p), . . . , xn(p)) is the set of coordinates of p w.r.t. the
chart.
The inverse, (Ω, ϕ−1), of a chart is called a

local parametrization.
Given any two charts, (U1, ϕ1) and (U2, ϕ2), if

U1 ∩ U2 6= ∅, we have the transition maps,
ϕji : ϕi(Ui ∩ Uj ) → ϕj (Ui ∩ Uj ) and
ϕij : ϕj (Ui ∩ Uj ) → ϕi(Ui ∩ Uj ), defined by
ϕji = ϕj ◦ ϕ−1
i and ϕ i
j = ϕ i ◦ ϕ −1
j .
6.1. MANIFOLDS 309
Clearly, ϕij = (ϕji )−1.
Observe that the transition maps ϕji (resp. ϕij ) are maps
between open subsets of Rn.
This is good news! Indeed, the whole arsenal of calculus

is available for functions on Rn, and we will be able to
promote many of these results to manifolds by imposing
suitable conditions on transition functions.
Definition 6.1.2 Given a topological space, M , and

any two integers, n ≥ 1 and k ≥ 1, a C k n-atlas (or
n-atlas of class C k ), A, is a family of charts, {(Ui, ϕi)},
such that
(1) ϕi(Ui) ⊆ Rn for all i;
(2) The Ui cover M , i.e.,
[
M= Ui ;
i
(3) Whenever Ui ∩ Uj 6= ∅, the transition map ϕji (and

ϕij ) is a C k -diffeomorphism.
We must insure that we have enough charts in order to

carry out our program of generalizing calculus on Rn to
manifolds.
For this, we must be able to add new charts whenever

necessary, provided that they are consistent with the pre-
vious charts in an existing atlas.
Technically, given a C k n-atlas, A, on M , for any other

chart, (U, ϕ), we say that (U, ϕ) is compatible with the
altas A iff every map ϕi ◦ϕ−1 and ϕ◦ϕ−1 k
i is C (whenever
U ∩ Ui 6= ∅).
Two atlases A and A0 on M are compatible iff every

chart of one is compatible with the other atlas.
This is equivalent to saying that the union of the two

atlases is still an atlas.
6.1. MANIFOLDS 311
It is immediately verified that compatibility induces an

equivalence relation on C k n-atlases on M .
In fact, given an atlas, A, for M , the collection, A,

e of
all charts compatible with A is a maximal atlas in the
equivalence class of charts compatible with A.
Definition 6.1.3 Given any two integers, n ≥ 1 and

k ≥ 1, a C k -manifold of dimension n consists of a topo-
logical space, M , together with an equivalence class, A,
of C k n-atlases, on M . Any atlas, A, in the equivalence
class A is called a differentiable structure of class C k
(and dimension n) on M . We say that M is modeled on
Rn. When k = ∞, we say that M is a smooth manifold .
Remark: It might have been better to use the terminol-

ogy abstract manifold rather than manifold, to empha-
size the fact that the space M is not a priori a subspace
of RN , for some suitable N .
We can allow k = 0 in the above definitions. Condition

(3) in Definition 6.1.2 is void, since a C 0-diffeomorphism
is just a homeomorphism, but ϕji is always a homeomor-
phism.
In this case, M is called a topological manifold of di-

mension n.
We do not require a manifold to be connected but we

require all the components to have the same dimension,
n.
Actually, on every connected component of M , it can be

shown that the dimension, nϕ, of the range of every chart
is the same. This is quite easy to show if k ≥ 1 but for
k = 0, this requires a deep theorem of Brouwer.
What happens if n = 0? In this case, every one-point

subset of M is open, so every subset of M is open, i.e., M
is any (countable if we assume M to be second-countable)
set with the discrete topology!
6.1. MANIFOLDS 313
Observe that since Rn is locally compact and locally con-

nected, so is every manifold.
Remark: In some cases, M does not come with a topol-
ogy in an obvious (or natural) way and a slight variation
of Definition 6.1.2 is more convenient in such a situation:
Definition 6.1.4 Given a set, M , and any two integers,

n ≥ 1 and k ≥ 1, a C k n-atlas (or n-atlas of class C k ),
A, is a family of charts, {(Ui, ϕi)}, such that
(1) Each Ui is a subset of M and ϕi: Ui → ϕi(Ui) is a
bijection onto an open subset, ϕi(Ui) ⊆ Rn, for all i;
(2) The Ui cover M , i.e.,
[
M= Ui ;
i
(3) Whenever Ui ∩ Uj 6= ∅, the set ϕi(Ui ∩ Uj ) is open

in Rn and the transition map ϕji (and ϕij ) is a C k -
diffeomorphism.
Then, the notion of a chart being compatible with an
atlas and of two atlases being compatible is just as before
and we get a new definition of a manifold, analogous to
Definition 6.1.3.
But, this time, we give M the topology in which the

open sets are arbitrary unions of domains of charts, Ui,
more precisely, the Ui’s of the maximal atlas defining the
differentiable structure on M .
It is not difficult to verify that the axioms of a topology

are verified and M is indeed a topological space with this
topology.
It can also be shown that when M is equipped with the

above topology, then the maps ϕi: Ui → ϕi(Ui) are home-
omorphisms, so M is a manifold according to Definition
6.1.3. Thus, we are back to the original notion of a man-
ifold where it is assumed that M is already a topological
space.
One can also define the topology on M in terms of any

the atlases, A, defining M (not only the maximal one) by
requiring U ⊆ M to be open iff ϕi(U ∩ Ui) is open in Rn,
for every chart, (Ui, ϕi), in the altas A. This topology is
the same as the topology induced by the maximal atlas.
We also require M to be Hausdorff and second-countable

with this topology. If M has a countable atlas, then it is
second-countable
6.1. MANIFOLDS 315
If the underlying topological space of a manifold is com-

pact, then M has some finite atlas.
Also, if A is some atlas for M and (U, ϕ) is a chart in A,

for any (nonempty) open subset, V ⊆ U , we get a chart,
(V, ϕ V ), and it is obvious that this chart is compatible
with A.
Thus, (V, ϕ V ) is also a chart for M . This observation

shows that if U is any open subset of a C k -manifold,
M , then U is also a C k -manifold whose charts are the
restrictions of charts on M to U .
Example 1. The sphere S n.

Using the stereographic projections (from the north pole
and the south pole), we can define two charts on S n and
show that S n is a smooth manifold. Let
σN : S n − {N } → Rn and σS : S n − {S} → Rn, where
N = (0, · · · , 0, 1) ∈ Rn+1 (the north pole) and
S = (0, · · · , 0, −1) ∈ Rn+1 (the south pole) be the maps
called respectively stereographic projection from the north
pole and stereographic projection from the south pole
given by
1
σN (x1, . . . , xn+1) = (x1, . . . , xn)
1 − xn+1
and
1
σS (x1, . . . , xn+1) = (x1, . . . , xn).
1 + xn+1
The inverse stereographic projections are given by
−1
σN (x1, . . . , xn) =
n
1 X
Pn 2 (2x1, . . . , 2xn, ( x2i ) − 1)
( i=1 xi ) + 1 i=1
and
σS−1(x1, . . . , xn) =
n
1 X
Pn 2 (2x1, . . . , 2xn, −( x2i ) + 1).
( i=1 xi ) + 1 i=1
Thus, if we let UN = S n − {N } and US = S n − {S},

we see that UN and US are two open subsets covering S n,
both homeomorphic to Rn.
6.1. MANIFOLDS 317
Furthermore, it is easily checked that on the overlap,

UN ∩ US = S n − {N, S}, the transition maps
−1
σS ◦ σN = σN ◦ σS−1
are given by
1
(x1, . . . , xn) 7→ n
P 2 (x1, . . . , xn),
x
i=1 i
that is, the inversion of center O = (0, . . . , 0) and power
1. Clearly, this map is smooth on Rn − {O}, so we con-
clude that (UN , σN ) and (US , σS ) form a smooth atlas for
S n.
Example 2. The projective space RPn.
To define an atlas on RPn it is convenient to view RPn

as the set of equivalence classes of vectors in Rn+1 − {0}
modulo the equivalence relation,
u ∼ v iff v = λu, for some λ 6= 0 ∈ R.
Given any p = [x1, . . . , xn+1] ∈ RPn, we call (x1, . . . , xn+1)
the homogeneous coordinates of p.
It is customary to write (x1: · · · : xn+1) instead of

[x1, . . . , xn+1]. (Actually, in most books, the indexing
starts with 0, i.e., homogeneous coordinates for RPn are
written as (x0: · · · : xn).)
For any i, with 1 ≤ i ≤ n + 1, let

Ui = {(x1: · · · : xn+1) ∈ RPn | xi 6= 0}.
Observe that Ui is well defined, because if
(y1: · · · : yn+1) = (x1: · · · : xn+1), then there is some λ 6= 0
so that yi = λxi, for i = 1, . . . , n + 1.
We can define a homeomorphism, ϕi, of Ui onto Rn, as

follows:

x1 xi−1 xi+1 xn+1
ϕi(x1: · · · : xn+1) = ,..., , ,..., ,
xi xi xi xi
where the ith component is omitted. Again, it is clear
that this map is well defined since it only involves ratios.
6.1. MANIFOLDS 319
We can also define the maps, ψi, from Rn to Ui ⊆ RPn,

given by
ψi(x1, . . . , xn) = (x1: · · · : xi−1: 1: xi: · · · : xn),
where the 1 goes in the ith slot, for i = 1, . . . , n + 1.
One easily checks that ϕi and ψi are mutual inverses, so

the ϕi are homeomorphisms. On the overlap, Ui ∩ Uj ,
(where i 6= j), as xj 6= 0, we have
(ϕj ◦ ϕ−1
i)(x1 , . . . , xn ) =
x1 xi−1 1 xi xj−1 xj+1 xn
,..., , , ,..., , ,..., .
xj xj xj xj xj xj xj
(We assumed that i < j; the case j < i is similar.) This is
clearly a smooth function from ϕi(Ui ∩Uj ) to ϕj (Ui ∩Uj ).
As the Ui cover RPn, see conclude that the (Ui, ϕi) are
n + 1 charts making a smooth atlas for RPn.
Intuitively, the space RPn is obtained by glueing the open

subsets Ui on their overlaps. Even for n = 3, this is not
easy to visualize!
Example 3. The Grassmannian G(k, n).
Recall that G(k, n) is the set of all k-dimensional linear

subspaces of Rn, also called k-planes.
Every k-plane, W , is the linear span of k linearly indepen-

dent vectors, u1, . . . , uk , in Rn; furthermore, u1, . . . , uk
and v1, . . . , vk both span W iff there is an invertible k×k-
matrix, Λ = (λij ), such that
k
X
vi = λij uj , 1 ≤ i ≤ k.
j=1
Obviously, there is a bijection between the collection of

k linearly independent vectors, u1, . . . , uk , in Rn and the
collection of n × k matrices of rank k.
Furthermore, two n × k matrices A and B of rank k

represent the same k-plane iff
B = AΛ, for some invertible k × k matrix, Λ.
6.1. MANIFOLDS 321
(Note the analogy with projective spaces where two vec-

tors u, v represent the same point iff v = λu for some
invertible λ ∈ R.)
We can define the domain of charts (according to Def-

inition 6.1.4) on G(k, n) as follows: For every subset,
S = {i1, . . . , ik } of {1, . . . , n}, let US be the subset of
n × k matrices, A, of rank k whose rows of index in
S = {i1, . . . , ik } forms an invertible k × k matrix de-
noted AS .
Observe that the k × k matrix consisting of the rows of

the matrix AA−1
S whose index belong to S is the identity
matrix, Ik .
Therefore, we can define a map, ϕS : US → R(n−k)×k ,

where ϕS (A) = the (n − k) × k matrix obtained by delet-
ing the rows of index in S from AA−1
S .
We need to check that this map is well defined, i.e., that

it does not depend on the matrix, A, representing W .
Let us do this in the case where S = {1, . . . , k}, which

is notationally simpler. The general case can be reduced
to this one using a suitable permutation.
If B = AΛ, with Λ invertible, if we write

A1 B1
A= and B = ,
A2 B2
as B = AΛ, we get B1 = A1Λ and B2 = A2Λ, from
which we deduce that

B1 Ik
B1−1 = −1 =
B2 B2B1

Ik Ik A1
−1 −1 = −1 = A−1
1 .
A2ΛΛ A1 A2 A1 A2
Therefore, our map is indeed well-defined.

6.1. MANIFOLDS 323
It is clearly injective and we can define its inverse, ψS , as

follows: Let πS be the permutation of {1, . . . , n} swaping
{1, . . . , k} and S and leaving every other element fixed
(i.e., if S = {i1, . . . , ik }, then πS (j) = ij and πS (ij ) = j,
for j = 1, . . . , k).
If PS is the permutation matrix associated with πS , for

any (n − k) × k matrix, M , let

Ik
ψS (M ) = PS .
M
The effect of ψS is to “insert into M ” the rows of the
identity matrix Ik as the rows of index from S.
At this stage, we have charts that are bijections from

subsets, US , of G(k, n) to open subsets, namely, R(n−k)×k .
Then, the reader can check that the transition map

ϕT ◦ ϕ−1
S from ϕS (US ∩ UU ) to ϕT (US ∩ UU ) is given by
M 7→ (C + DM )(A + BM )−1,
where
A B
= PT PS ,
C D
is the matrix of the permutation πT ◦πS (this permutation
“shuffles” S and T ).
This map is smooth, as it is given by determinants, and

so, the charts (US , ϕS ) form a smooth atlas for G(k, n).
Finally, one can check that the conditions of Definition

6.1.4 are satisfied, so the atlas just defined makes G(k, n)
into a topological space and a smooth manifold.
Remark: The reader should have no difficulty proving
that the collection of k-planes represented by matrices
in US is precisely set of k-planes, W , supplementary to
the (n − k)-plane spanned by the n − k canonical basis
vectors ejk+1 , . . . , ejn (i.e., span(W ∪ {ejk+1 , . . . , ejn }) =
Rn, where S = {i1, . . . , ik } and
{jk+1, . . . , jn} = {1, . . . , n} − S).
6.1. MANIFOLDS 325
Example 4. Product Manifolds.
Let M1 and M2 be two C k -manifolds of dimension n1 and

n2, respectively.
The topological space, M1 × M2, with the product topol-

ogy (the opens of M1 × M2 are arbitrary unions of sets of
the form U × V , where U is open in M1 and V is open
in M2) can be given the structure of a C k -manifold of
dimension n1 + n2 by defining charts as follows:
For any two charts, (Ui, ϕi) on M1 and (Vj , ψj ) on M2,

we declare that (Ui × Vj , ϕi × ψj ) is a chart on M1 × M2,
where ϕi × ψj : Ui × Vj → Rn1+n2 is defined so that
ϕi × ψj (p, q) = (ϕi(p), ψj (q)), for all (p, q) ∈ Ui × Vj .
We define C k -maps between manifolds as follows:

Definition 6.1.5 Given any two C k -manifolds, M and

N , of dimension m and n respectively, a C k -map if a
continuous functions, h: M → N , so that for every
p ∈ M , there is some chart, (U, ϕ), at p and some chart,
(V, ψ), at q = h(p), with f (U ) ⊆ V and
ψ ◦ h ◦ ϕ−1: ϕ(U ) −→ ψ(V )
a C k -function.
It is easily shown that Definition 6.1.5 does not depend on

the choice of charts. In particular, if N = R, we obtain
a C k -function on M .
One checks immediately that a function, f : M → R, is a

C k -map iff for every p ∈ M , there is some chart, (U, ϕ),
at p so that
f ◦ ϕ−1: ϕ(U ) −→ R
is a C k -function.
6.1. MANIFOLDS 327
If U is an open subset of M , set of C k -functions on U is

denoted by C k (U ). In particular, C k (M ) denotes the set
of C k -functions on the manifold, M . Observe that C k (U )
is a ring.
On the other hand, if M is an open interval of R, say

M =]a, b[ , then γ: ]a, b[ → N is called a C k -curve in N .
One checks immediately that a function, γ: ]a, b[ → N , is
a C k -map iff for every q ∈ N , there is some chart, (V, ψ),
at q so that
ψ ◦ γ: ]a, b[ −→ ψ(V )
is a C k -function.
It is clear that the composition of C k -maps is a C k -map.

A C k -map, h: M → N , between two manifolds is a C k -
diffeomorphism iff h has an inverse, h−1: N → M (i.e.,
h−1 ◦h = idM and h◦h−1 = idN ), and both h and h−1 are
C k -maps (in particular, h and h−1 are homeomorphisms).
Next, we define tangent vectors.
6.2 Tangent Vectors, Tangent Spaces,

Cotangent Spaces
Let M be a C k manifold of dimension n, with k ≥ 1.

The most intuitive method to define tangent vectors is to
use curves. Let p ∈ M be any point on M and let
γ: ] − , [ → M be a C 1-curve passing through p, that
is, with γ(0) = p. Unfortunately, if M is not embed-
ded in any RN , the derivative γ 0(0) does not make sense.
However, for any chart, (U, ϕ), at p, the map ϕ ◦ γ is a
C 1-curve in Rn and the tangent vector v = (ϕ ◦ γ)0(0)
is well defined. The trouble is that different curves may
yield the same v!
To remedy this problem, we define an equivalence relation

on curves through p as follows:
6.2. TANGENT VECTORS, TANGENT SPACES, COTANGENT SPACES 329
Definition 6.2.1 Given a C k manifold, M , of dimen-

sion n, for any p ∈ M , two C 1-curves, γ1: ] − 1, 1[ → M
and γ2: ]−2, 2[→ M , through p (i.e., γ1(0) = γ2(0) = p)
are equivalent iff there is some chart, (U, ϕ), at p so that
(ϕ ◦ γ1)0(0) = (ϕ ◦ γ2)0(0).
Now, the problem is that this definition seems to depend

on the choice of the chart. Fortunately, this is not the
case.
This leads us to the first definition of a tangent vector.
Definition 6.2.2 (Tangent Vectors, Version 1) Given

any C k -manifold, M , of dimension n, with k ≥ 1, for any
p ∈ M , a tangent vector to M at p is any equivalence
class of C 1-curves through p on M , modulo the equiva-
lence relation defined in Definition 6.2.1. The set of all
tangent vectors at p is denoted by Tp(M ).
It is obvious that Tp(M ) is a vector space.
We will show that Tp(M ) is a vector space of dimension

n = dimension of M .
One should observe that unless M = Rn, in which case,

for any p, q ∈ Rn, the tangent space Tq (M ) is naturally
isomorphic to the tangent space Tp(M ) by the translation
q − p, for an arbitrary manifold, there is no relationship
between Tp(M ) and Tq (M ) when p 6= q.
One of the defects of the above definition of a tangent

vector is that it has no clear relation to the C k -differential
structure of M .
In particular, the definition does not seem to have any-

thing to do with the functions defined locally at p.
There is another way to define tangent vectors that re-

veals this connection more clearly. Moreover, such a def-
inition is more intrinsic, i.e., does not refer explicitly to
charts.
As a first step, consider the following: Let (U, ϕ) be a

chart at p ∈ M (where M is a C k -manifold of dimen-
sion n, with k ≥ 1) and let xi = pri ◦ ϕ, the ith local
coordinate (1 ≤ i ≤ n).
For any function, f , defined on U 3 p, set

−1

∂ ∂(f ◦ ϕ )
f= , 1 ≤ i ≤ n.
∂xi p ∂Xi ϕ(p)
(Here, (∂g/∂Xi)|y denotes the partial derivative of a func-
tion g: Rn → R with respect to the ith coordinate, eval-
uated at y.)
We would expect that the function that maps f to the

above value is a linear map on the set of functions defined
locally at p, but there is technical difficulty:
The set of functions defined locally at p is not a vector

space!
To see this, observe that if f is defined on an open U 3 p

and g is defined on a different open V 3 p, then we do
know how to define f + g.
The problem is that we need to identify functions that

agree on a smaller open. This leads to the notion of
germs.
Definition 6.2.3 Given any C k -manifold, M , of dimen-

sion n, with k ≥ 1, for any p ∈ M , a locally defined
function at p is a pair, (U, f ), where U is an open sub-
set of M containing p and f is a function defined on U .
Two locally defined functions, (U, f ) and (V, g), at p are
equivalent iff there is some open subset, W ⊆ U ∩ V ,
containing p so that
f W = g W.
The equivalence class of a locally defined function at p,
denoted [f ] or f , is called a germ at p.
One should check that the relation of Definition 6.2.3 is

indeed an equivalence relation.
Of course, the value at p of all the functions, f , in any

germ, f , is f (p). Thus, we set f (p) = f (p).
One should also check that we can define addition of

germs, multiplication of a germ by a scalar and multi-
plication of germs, in the obvious way:
If f and g are two germs at p, and λ ∈ R, then

[f ] + [g] = [f + g]
λ[f ] = [λf ]
[f ][g] = [f g].
(Of course, f + g is the function locally defined so that
(f + g)(x) = f (x) + g(x) and similarly, (λf )(x) = λf (x)
and (f g)(x) = f (x)g(x).)
Therefore, the germs at p form a ring. The ring of germs

(k)
of C k -functions at p is denoted OM,p. When k = ∞, we
usually drop the superscript ∞.
Remark: Most readers will most likely be puzzled by
(k)
the notation OM,p.
In fact, it is standard in algebraic geometry, but it is not

as commonly used in differential geometry.
For any open subset, U , of a manifold, M , the ring,

(k)
C k (U ), of C k -functions on U is also denoted OM (U ) (cer-
tainly by people with an algebraic geometry bent!).
(k)
Then, it turns out that the map U 7→ OM (U ) is a sheaf ,
(k) (k)
denoted OM , and the ring OM,p is the stalk of the sheaf
(k)
OM at p.
Such rings are called local rings. Roughly speaking, all

the “local” information about M at p is contained in the
(k)
local ring OM,p. (This is to be taken with a grain of
salt. In the C k -case where k < ∞, we also need the
“stationary germs”, as we will see shortly.)
Now that we have a rigorous way of dealing with functions

locally defined at p, observe that the map

∂
vi: f 7→ f
∂xi p
yields the same value for all functions f in a germ f at p.
(k)
Furthermore, the above map is linear on OM,p. More is
true.
Firstly for any two functions f, g locally defined at p, we

have

∂ ∂ ∂
(f g) = f (p) g + g(p) f.
∂xi p ∂xi p ∂xi p
Secondly, if (f ◦ ϕ−1)0(ϕ(p)) = 0, then

∂
f = 0.
∂xi p
The first property says that vi is a derivation. As to the

second property, when (f ◦ ϕ−1)0(ϕ(p)) = 0, we say that
f is stationary at p.
It is easy to check (using the chain rule) that being sta-

tionary at p does not depend on the chart, (U, ϕ), at p
or on the function chosen in a germ, f . Therefore, the
notion of a stationary germ makes sense:
We say that f is a stationary germ iff

(f ◦ ϕ−1)0(ϕ(p)) = 0 for some chart, (U, ϕ), at p and
some function, f , in the germ, f .
(k)
The C k -stationary germs form a subring of OM,p (but not
(k)
an ideal!) denoted SM,p.
Remarkably, it turns out that the dual of the vector space,

(k) (k)
OM,p/SM,p, is isomorphic to the tangent space, Tp(M ).
(k)
First, we prove that the subspace of linear forms on OM,p

(k)
that vanish on SM,p has ∂x∂ 1 , . . . , ∂x∂ n as a basis.
p p
Proposition 6.2.4 Given any C k -manifold, M , of di-

mension n, with k ≥ 1, for any
p∈ M and any chart
(U, ϕ) at p, the n functions, ∂x∂ 1 , . . . , ∂x∂ n , de-
p p
(k)
fined on OM,p by
−1

∂ ∂(f ◦ ϕ )
f= , 1≤i≤n
∂xi p ∂Xi
ϕ(p)
(k)
are linear forms that vanish on SM,p. Every linear
(k) (k)
form, L, on OM,p that vanishes on SM,p can be ex-
pressed in a unique way as
n
X ∂
L= λi ,
i=1
∂xi p
where λi ∈ R. Therefore, the

∂
, i = 1, . . . , n
∂xi p
form a basis of the vector space of linear forms on
(k) (k)
OM,p that vanish on SM,p.
(k)
As the subspace of linear forms on OM,p that vanish on
(k) (k) (k)
SM,p is isomorphic to the dual, (OM,p/SM,p)∗, of the space
(k) (k)
OM,p/SM,p, we see that the

∂
, i = 1, . . . , n
∂xi p
(k) (k)
also form a basis of (OM,p/SM,p)∗.
To define our second version of tangent vectors, we need

to define linear derivations.

sion n, with k ≥ 1, for any p ∈ M , a linear derivation
(k)
at p is a linear form, v, on OM,p, such that
v(fg) = f (p)v(g) + g(p)v(f ),
(k)
for all germs f , g ∈ OM,p. The above is called the Leib-
nitz property.

∂
Recall that we observed earlier that the ∂xi p are linear
derivations at p. Therefore, we have
Proposition 6.2.6 Given any C k -manifold, M , of di-

mension n, with k ≥ 1, for any p ∈ M , the linear
(k) (k)
forms on OM,p that vanish on SM,p are exactly the
(k) (k)
linear derivations on OM,p that vanish on SM,p.
Here is now our second definition of a tangent vector.

any C k -manifold, M , of dimension n, with k ≥ 1, for
any p ∈ M , a tangent vector to M at p is any linear
(k) (k)
derivation on OM,p that vanishes on SM,p, the subspace
of stationary germs.
Let us consider the simple case where M = R. In this

case, for every x ∈ R, the tangent space, Tx(R), is a one-
dimensional vector space isomorphic to R and
∂ d

∂t x = dt x is a basis vector of Tx (R).
For every C k -function, f , locally defined at x, we have

∂ df
f = = f 0(x).
∂t x dt x
∂

Thus, ∂t x is: compute the derivative of a function at x.
We now prove the equivalence of the two Definitions of a

tangent vector.
Proposition 6.2.8 Let M be any C k -manifold of di-

mension n, with k ≥ 1. For any p ∈ M , let u be any
tangent vector (version 1) given by some equivalence
class of C 1-curves, γ: ] − , +[ → M , through p (i.e.,
(k)
p = γ(0)). Then, the map Lu defined on OM,p by
Lu(f ) = (f ◦ γ)0(0)
(k)
is a linear derivation that vanishes on SM,p. Further-
more, the map u 7→ Lu defined above is an isomor-
(k) (k)
phism between Tp(M ) and (OM,p/SM,p)∗, the space of
(k) (k)
linear forms on OM,p that vanish on SM,p.
In view of Proposition 6.2.8, we can identify Tp(M ) with

(k) (k)
(OM,p/SM,p)∗.
(k) (k)
As the space OM,p/SM,p is finite dimensional,
(k) (k) (k) (k)
(OM,p/SM,p)∗∗ is canonically isomorphic to OM,p/SM,p,
(k) (k)
so we can identify Tp∗(M ) with OM,p/SM,p.
(Recall that if E is a finite dimensional space, the map

iE : E → E ∗∗ defined so that, for any v ∈ E,
v 7→ ve, where ve(f ) = f (v), for all f ∈ E ∗
is a linear isomorphism.) This also suggests the following
definition:

sion n, with k ≥ 1, for any p ∈ M , the tangent space
at p, denoted Tp(M ), is the space of linear derivations on
(k) (k)
OM,p that vanish on SM,p. Thus, Tp(M ) can be identi-
(k) (k) (k) (k)
fied with (OM,p/SM,p)∗. The space OM,p/SM,p is called
the cotangent space at p; it is isomorphic to the dual,
Tp∗(M ), of Tp(M ).
Observe that if xi = pri ◦ ϕ, as

∂
xj = δi,j ,
∂xi p
(k) (k)
the images of x1, . . . , xn in OM,p/SM,p are the dual of the

basis ∂x1 , . . . , ∂x∂ n of Tp(M ).
∂
p p
Given any C k -function, f , on M , we denote the image of

(k) (k)
f in Tp∗(M ) = OM,p/SM,p by dfp.
This is the differential of f at p.

(k) (k)
Using the isomorphism between OM,p/SM,p and
(k) (k)
(OM,p/SM,p)∗∗ described above, dfp corresponds to the
linear map in Tp∗(M ) defined by dfp(v) = v(f ), for all
v ∈ Tp(M ).
With this notation, we see that (dx1)p, . . . , (dxn)p is a

∗
basis
of Tp (M
), and
this basis is dual to the basis
∂ ∂
∂x1 p , . . . , ∂xn p of Tp(M ).
For simplicity of notation, we often omit the subscript p

unless confusion arises.
Remark: Strictly speaking, a tangent vector, v ∈ Tp(M ),
(k)
is defined on the space of germs, OM,p at p. However, it is
often convenient to define v on C k -functions f ∈ C k (U ),
where U is some open subset containing p. This is easy:
Set
v(f ) = v(f ).
Given any chart, (U, ϕ), at p, since v can be written in a
unique way as
n
X ∂
v= λi ,
i=1
∂x i p
we get
n
X ∂
v(f ) = λi f.
i=1
∂xi p
This shows that v(f ) is the directional derivative of f

in the direction v.
When M is a smooth manifold, things get a little sim-

pler. Indeed, it turns out that in this case, every linear
derivation vanishes on stationary germs.
To prove this, we recall the following result from calculus

(see Warner [?]):
Proposition 6.2.10 If g: Rn → R is a C k -function

(k ≥ 2) on a convex open, U , about p ∈ Rn, then for
every q ∈ U , we have
n
X ∂g
g(q) = g(p) + (qi − pi)
i=1
∂X
i p
n Z 1 2

X ∂ g
+ (qi − pi)(qj − pj ) (1 − t) dt.
i,j=1 0 ∂Xi∂Xj (1−t)p+tq
In particular, if g ∈ C ∞(U ), then the integral as a

function of q is C ∞.
Proposition 6.2.11 Let M be any C ∞-manifold of

dimension n. For any p ∈ M , any linear derivation
(∞)
on OM,p vanishes on stationary germs.
Proposition 6.2.11 shows that in the case of a smooth

manifold, in Definition 6.2.7, we can omit the requirement
that linear derivations vanish on stationary germs, since
this is automatic.
(∞)
It is also possible to define Tp(M ) just in terms of OM,p .
(∞)
Let mM,p ⊆ OM,p be the ideal of germs that vanish at
p. Then, we also have the ideal m2M,p, which consists of
all finite sums of products of two elements in mM,p, and
it can be shown that Tp∗(M ) is isomorphic to mM,p/m2M,p
(see Warner [?], Lemma 1.16).
(k)
Actually, if we let mM,p denote the C k germs that vanish
(k)
at p and sM,p denote the stationary C k -germs that vanish
at p, it is easy to show that
OM,p/SM,p ∼
(k) (k) (k) (k)
= mM,p/sM,p.
(k) (k)
(Given any f ∈ OM,p, send it to f − f (p) ∈ mM,p.)
(k)
Clearly, (mM,p)2 consists of stationary germs (by the deriva-
tion property) and when k = ∞, Proposition 6.2.10
shows that every stationary germ that vanishes at p be-
longs to m2M,p. Therefore, when k = ∞, we have
(∞)
sM,p = m2M,p and so,
Tp∗(M ) = OM,p /SM,p ∼

(∞) (∞)
= mM,p/m2M,p.
(k)
Remark: The ideal mM,p is in fact the unique maximal
(k)
ideal of OM,p.
(k)
Thus, OM,p is a local ring (in the sense of commutative
algebra) called the local ring of germs of C k -functions
at p. These rings play a crucial role in algebraic geometry.
Yet one more way of defining tangent vectors will make

it a little easier to define tangent bundles.

any C k -manifold, M , of dimension n, with k ≥ 1, for any
p ∈ M , consider the triples, (U, ϕ, u), where (U, ϕ) is any
chart at p and u is any vector in Rn. Say that two such
triples (U, ϕ, u) and (V, ψ, v) are equivalent iff
(ψ ◦ ϕ−1)0ϕ(p)(u) = v.
A tangent vector to M at p is an equivalence class of
triples, [(U, ϕ, u)], for the above equivalence relation.
The intuition behind Definition 6.2.12 is quite clear: The

vector u is considered as a tangent vector to Rn at ϕ(p).
If (U, ϕ) is a chart on M at p, we can define a natural iso-

morphism, θU,ϕ,p: Rn → Tp(M ), between Rn and Tp(M ),
as follows: For any u ∈ Rn,
θU,ϕ,p: u 7→ [(U, ϕ, u)].
One immediately check that the above map is indeed lin-
ear and a bijection.
The equivalence of this definition with the definition in

terms of curves (Definition 6.2.2) is easy to prove.
Proposition 6.2.13 Let M be any C k -manifold of

dimension n, with k ≥ 1. For any p ∈ M , let x be any
tangent vector (version 1) given by some equivalence
class of C 1-curves, γ: ] − , +[ → M , through p (i.e.,
p = γ(0)). The map
x 7→ [(U, ϕ, (ϕ ◦ γ)0(0))]
is an isomorphism between Tp(M )-version 1 and Tp(M )-
version 3.
For simplicity of notation, we also use the notation TpM

for Tp(M ) (resp. Tp∗M for Tp∗(M )).
After having explored thorougly the notion of tangent

vector, we show how a C k -map, h: M → N , between C k
manifolds, induces a linear map, dhp: Tp(M ) → Th(p)(N ),
for every p ∈ M .
We find it convenient to use Version 2 of the definition of

a tangent vector. So, let u ∈ Tp(M ) be a linear derivation
(k) (k)
on OM,p that vanishes on SM,p.
(k)
We would like dhp(u) to be a linear derivation on ON,h(p)
(k)
that vanishes on SN,h(p).
(k)
So, for every germ, g ∈ ON,h(p), set
dhp(u)(g) = u(g ◦ h).
For any locally defined function, g, at h(p) in the germ,

g (at h(p)), it is clear that g ◦ h is locally defined at p
and is C k , so g ◦ h is indeed a C k -germ at p.
Moreover, if g is a stationary germ at h(p), then for some

chart, (V, ψ) on N at q = h(p), we have
(g ◦ ψ −1)0(ψ(q)) = 0 and, for some chart (U, ϕ) at p on
M , we get
(g◦h◦ϕ−1)0(ϕ(p)) = (g◦ψ −1)(ψ(q))((ψ◦h◦ϕ−1)0(ϕ(p)))
= 0,
which means that g ◦ h is stationary at p.
Therefore, dhp(u) ∈ Th(p)(M ). It is also clear that dhp

is a linear map. We summarize all this in the following
definition:
Definition 6.2.14 Given any two C k -manifolds, M and

N , of dimension m and n, respectively, for any C k -map,
h: M → N , and for every p ∈ M , the differential of
h at p or tangent map, dhp: Tp(M ) → Th(p)(N ), is the
linear map defined so that
dhp(u)(g) = u(g ◦ h),
(k)
for every u ∈ Tp(M ) and every germ, g ∈ ON,h(p). The
linear map dhp is also denoted Tph (and sometimes, h0p
or Dph).
The chain rule is easily generalized to manifolds.
Proposition 6.2.15 Given any two C k -maps

f : M → N and g: N → P between smooth C k -manifolds,
for any p ∈ M , we have
d(g ◦ f )p = dgf (p) ◦ dfp.
In the special case where N = R, a C k -map between the

manifolds M and R is just a C k -function on M .
It is interesting to see what dfp is explicitly. Since N = R,

germs (of functions on R) at t0 = f (p) are just germs of
C k -functions, g: R → R, locally defined at t0.
Then, for any u ∈ Tp(M ) and every germ g at t0,

dfp(u)(g) = u(g ◦ f ).
If
wepick a chart, (U, ϕ), on M at p, we know that the
∂
∂xi form a basis of Tp(M ), with 1 ≤ i ≤ n.
p
Therefore, it is enough to figure out what dfp(u)(g) is

when u = ∂x∂ i . In this case,
p
! −1

∂ ∂(g ◦ f ◦ ϕ )
dfp (g) = .
∂xi p ∂Xi
ϕ(p)
Using the chain rule, we find that

!
∂ ∂ dg
dfp (g) = f .
∂xi p ∂xi p dt t0
Therefore, we have

d
dfp(u) = u(f ) .
dt t0
This shows that we can identify dfp with the linear form
in Tp∗(M ) defined by
dfp(v) = v(f ).
This is consistent with our previous definition of dfp as
(k) (k)
the image of f in Tp∗(M ) = OM,p/SM,p (as Tp(M ) is
(k) (k)
isomorphic to (OM,p/SM,p)∗).
In preparation for the definition of the flow of a vector

field (which will be needed to define the exponential map
in Lie group theory), we need to define the tangent vector
to a curve on a manifold.
Given a C k -curve, γ: ]a, b[ → M , on a C k -manifold, M ,

for any t0 ∈]a, b[ , we would like to define the tangent
vector to the curve γ at t0 as a tangent vector to M at
p = γ(t0).
d

We do this as follows: Recall that dt t0 is a basis vector
of Tt0 (R) = R.
So, define the tangent vector to the curve γ at t, denoted

γ̇(t0) (or γ 0(t), or dγ
dt (t0 )) by
!
d
γ̇(t) = dγt .
dt t0

Sometime, it is necessary to define curves (in a manifold)

whose domain is not an open interval.
A map, γ: [a, b] → M , is a C k -curve in M if it is the

restriction of some C k -curve, γ
e: ]a − , b + [ → M , for
some (small) > 0.
Note that for such a curve (if k ≥ 1) the tangent vector,

γ̇(t), is defined for all t ∈ [a, b],
A curve, γ: [a, b] → M , is piecewise C k iff there a se-

quence, a0 = a, a1, . . . , am = b, so that the restriction of
γ to each [ai, ai+1] is a C k -curve, for i = 0, . . . , m − 1.
6.3 Tangent and Cotangent Bundles, Vector Fields
Let M be a C k -manifold (with k ≥ 2). Roughly speaking,

a vector field on M is the assignment, p 7→ ξ(p), of a
tangent vector, ξ(p) ∈ Tp(M ), to a point p ∈ M .
Generally, we would like such assignments to have some

smoothness properties when p varies in M , for example,
to be C l , for some l related to k.
Now, if the collection, T (M ), of all tangent spaces, Tp(M ),

was a C l -manifold, then it would be very easy to define
what we mean by a C l -vector field: We would simply
require the maps, ξ: M → T (M ), to be C l .
If M is a C k -manifold of dimension n, then we can indeed

define make T (M ) into a C k−1-manifold of dimension 2n
and we now sketch this construction.
6.3. TANGENT AND COTANGENT BUNDLES, VECTOR FIELDS 357
We find it most convenient to use Version 3 of the def-

inition of tangent vectors, i.e., as equivalence classes of
triple (U, ϕ, u).
First, we let T (M ) be the disjoint union of the tangent

spaces Tp(M ), for all p ∈ M . There is a natural projec-
tion,
π: T (M ) → M, where π(v) = p if v ∈ Tp(M ).
We still have to give T (M ) a topology and to define a

C k−1-atlas.
For every chart, (U, ϕ), of M (with U open in M ) we

e π −1(U ) → R2n by
define the function ϕ:
−1
e = (ϕ ◦ π(v), θU,ϕ,π(v)
ϕ(v) (v)),
where v ∈ π −1(U ) and θU,ϕ,p is the isomorphism between
Rn and Tp(M ) described just after Definition 6.2.12.
e is a bijection between π −1(U ) and

It is obvious that ϕ
ϕ(U ) × Rn, an open subset of R2n.
We give T (M ) the weakest topology that makes all the

ϕ
e continuous, i.e., we take the collection of subsets of the
form ϕe−1(W ), where W is any open subset of ϕ(U )×Rn,
as a basis of the topology of T (M ).
One easily checks that T (M ) is Hausdorff and second-

countable in this topology. If (U, ϕ) and (V, ψ) are over-
lapping charts, then the transition function
e−1: ϕ(U ∩ V ) × Rn −→ ψ(U ∩ V ) × Rn
ψe ◦ ϕ
is given by
e−1(p, u) = (ψ ◦ ϕ−1(p), (ψ ◦ ϕ−1)0(u)).
ψe ◦ ϕ
e−1 is a C k−1-map. Therefore, T (M )
It is clear that ψe ◦ ϕ
is indeed a C k−1-manifold of dimension 2n, called the
tangent bundle.
Remark: Even if the manifold M is naturally embed-

ded in RN (for some N ≥ n = dim(M )), it is not at all
obvious how to view the tangent bundle, T (M ), as em-
0
bedded in RN , for sone suitable N 0. Hence, we see that
the definition of an abtract manifold is unavoidable.
A similar construction can be carried out for the cotan-

gent bundle.
In this case, we let T ∗(M ) be the disjoint union of the

cotangent spaces Tp∗(M ).
We also have a natural projection, π: T ∗(M ) → M , and

we can define charts as follows: For any chart, (U, ϕ), on
e π −1(U ) → R2n by
M , we define the function ϕ:
ϕ(τ
e )= ! !!

∂ ∂
ϕ ◦ π(τ ), τ ,...,τ ,
∂x1 π(τ ) ∂xn π(τ )

where τ ∈ π −1(U ) and the ∂
∂xi p are the basis of Tp(M )
associated with the chart (U, ϕ).
Again, one can make T ∗(M ) into a C k−1-manifold of di-

mension 2n, called the cotangent bundle.
Observe that for every chart, (U, ϕ), on M , there is a

bijection
τU : π −1(U ) → U × Rn,
given by
−1
τU (v) = (π(v), θU,ϕ,π(v) (v)).
Clearly, pr1 ◦ τU = π, on π −1(U ).
Thus, locally, that is, over U , the bundle T (M ) looks like

the product U × Rn.
We say that T (M ) is locally trivial (over U ) and we call

τU a trivializing map.
For any p ∈ M , the vector space π −1(p) = Tp(M ) is

called the fibre above p.
Observe that the restriction of τU to π −1(p) is an iso-

morphism between Tp(M ) and {p} × Rn ∼ = Rn, for any
p ∈ M.
All these ingredients are part of being a vector bundle

(but a little more is required of the maps τU ). For more
on bundles, see Lang [?], Gallot, Hulin and Lafontaine
[?], Lafontaine [?] or Bott and Tu [?].
When M = Rn, observe that

T (M ) = M × Rn = Rn × Rn, i.e., the bundle T (M ) is
(globally) trivial.
Given a C k -map, h: M → N , between two C k -manifolds,

we can define the function, dh: T (M ) → T (N ), (also
denoted T h, or h∗, or Dh) by setting
dh(u) = dhp(u), iff u ∈ Tp(M ).
We leave the next proposition as an exercise to the reader

(A proof can be found in Berger and Gostiaux [?]).
Proposition 6.3.1 Given a C k -map, h: M → N , be-

tween two C k -manifolds M and N (with k ≥ 1), the
map dh: T (M ) → T (N ) is a C k−1-map.
We are now ready to define vector fields.

Definition 6.3.2 Let M be a C k+1 manifold, with
k ≥ 1. For any open subset, U of M , a vector field on
U is any section, ξ, of T (M ) over U , i.e., any function,
ξ: U → T (M ), such that π ◦ ξ = idU (i.e., ξ(p) ∈ Tp(M ),
for every p ∈ U ). We also say that ξ is a lifting of U
into T (M ).
We say that ξ is a C h-vector field on U iff ξ is a section

over U and a C h-map, where 0 ≤ h ≤ k.
The set of C k -vector fields over U is denoted Γ(k)(U, T (M )).

Given a curve, γ: [a, b] → M , a vector field, ξ, along
γ is any section of T (M ) over γ, i.e., a C k -function,
ξ: [a, b] → T (M ), such that π ◦ ξ = γ. We also say
that ξ lifts γ into T (M ).
The above definition gives a precise meaning to the idea

that a C k -vector field on M is an assignment, p 7→ ξ(p),
of a tangent vector, ξ(p) ∈ Tp(M ), to a point, p ∈ M , so
that ξ(p) varies in a C k -fashion in terms of p.
Clearly, Γ(k)(U, T (M )) is a real vector space. For short,

the space Γ(k)(M, T (M )) is also denoted by Γ(k)(T (M ))
(or X(k)(M ) or even Γ(T (M )) or X(M )).
If M = Rn and U is an open subset of M , then

T (M ) = Rn ×Rn and a section of T (M ) over U is simply
a function, ξ, such that
ξ(p) = (p, u), with u ∈ Rn,
for all p ∈ U . In other words, ξ is defined by a function,
f : U → Rn (namely, f (p) = u).
This corresponds to the “old” definition of a vector field

in the more basic case where the manifold, M , is just Rn.
Given any C k -function, f ∈ C k (U ), and a vector field,

ξ ∈ Γ(k)(U, T (M )), we define the vector field, f ξ, by
(f ξ)(p) = f (p)ξ(p), p ∈ U.
Obviously, f ξ ∈ Γ(k)(U, T (M )), which shows that

Γ(k)(U, T (M )) is also a C k (U )-module. We also denote
ξ(p) by ξp.
For any chart, (U, ϕ), on M it is easy to check that the

map
∂
p 7→ , p ∈ U,
∂xi p
is a C k -vector field
on U (with 1 ≤ i ≤ n). This vector
field is denoted ∂x∂ i or ∂x∂ i .
If U is any open subset of M and f is any function in

C k (U ), then ξ(f ) is the function on U given by
ξ(f )(p) = ξp(f ) = ξp(f ).
As a special case, when (U, ϕ) is a chart on M , the vector

field, ∂x∂ i , just defined above induces the function

∂
p 7→ f, p ∈ U,
∂xi p

denoted ∂x∂ i (f ) or ∂x∂ i f .
It is easy to check that ξ(f ) ∈ C k−1(U ). As a conse-

quence, every vector field ξ ∈ Γ(k)(U, T (M )) induces a
linear map,
Lξ : C k (U ) −→ C k−1(U ),
given by f 7→ ξ(f ).
It is immediate to check that Lξ has the Leibnitz property,

i.e.,
Lξ (f g) = Lξ (f )g + f Lξ (g).
Linear maps with this property are called derivations.
Thus, we see that every vector field induces some kind of

differential operator, namely, a linear derivation.
Unfortunately, not every linear derivation of the above

type arises from a vector field, although this turns out to
be true in the smooth case i.e., when k = ∞ (for a proof,
see Gallot, Hulin and Lafontaine [?] or Lafontaine [?]).
In the rest of this section, unless stated otherwise, we

assume that k ≥ 1. The following easy proposition holds
(c.f. Warner [?]):
Proposition 6.3.3 Let ξ be a vector field on the C k+1-

manifold, M , of dimension n. Then, the following are
equivalent:
(a) ξ is C k .
(b) If (U, ϕ) is a chart on M and if f1, . . . , fn are the
functions on U uniquely defined by
n
X ∂
ξU = fi ,
i=1
∂xi
then each fi is a C k -map.

(c) Whenever U is open in M and f ∈ C k (U ), then
ξ(f ) ∈ C k−1(U ).
Given any two C k -vector field, ξ, η, on M , for any func-

tion, f ∈ C k (M ), we defined above the function ξ(f ) and
η(f ).
Thus, we can form ξ(η(f )) (resp. η(ξ(f ))), which are in

C k−2(M ).
Unfortunately, even in the smooth case, there is generally

no vector field, ζ, such that
ζ(f ) = ξ(η(f )), for all f ∈ C k (M ).
This is because ξ(η(f )) (and η(ξ(f ))) involve second-
order derivatives.
However, if we consider ξ(η(f )) − η(ξ(f )), then second-

order derivatives cancel out and there is a unique vector
field inducing the above differential operator.
Intuitively, ξη − ηξ measures the “failure of ξ and η to

commute”.
Proposition 6.3.4 Given any C k+1-manifold, M , of

dimension n, for any two C k -vector fields, ξ, η, on M ,
there is a unique C k−1-vector field, [ξ, η], such that
[ξ, η](f ) = ξ(η(f )) − η(ξ(f )), for all f ∈ C k−1(M ).
Definition 6.3.5 Given any C k+1-manifold, M , of di-

mension n, for any two C k -vector fields, ξ, η, on M , the
Lie bracket, [ξ, η], of ξ and η, is the C k−1 vector field
defined so that
[ξ, η](f ) = ξ(η(f )) − η(ξ(f )), for all f ∈ C k−1(M ).
We also have the following simple proposition whose proof

is left as an exercise (or, see Do Carmo [?]):
Proposition 6.3.6 Given any C k+1-manifold, M , of

dimension n, for any C k -vector fields, ξ, η, ζ, on M ,
for all f, g ∈ C k (M ), we have:
(a) [[ξ, η], ζ] + [[η, ζ], ξ] + [[ζ, ξ], η] = 0 (Jacobi iden-
tity).
(b) [ξ, ξ] = 0.
(c) [f ξ, gη] = f g[ξ, η] + f ξ(g)η − gη(f )ξ.
(d) [−, −] is bilinear.
Consequently, for smooth manifolds (k = ∞), the space

of vector fields, Γ(∞)(T (M )), is a vector space equipped
with a bilinear operation, [−, −], that satisfies the Jacobi
identity.
This makes Γ(∞)(T (M )) a Lie algebra.
One more notion will be needed when we deal with Lie

algebras.
Definition 6.3.7 Let ϕ: M → N be a C k+1-map of

manifolds. If ξ is a C k vector field on M and η is a C k
vector field on N , we say that ξ and η are ϕ-related iff
dϕ ◦ ξ = η ◦ ϕ.
Proposition 6.3.8 Let ϕ: M → N be a C k+1-map of

manifolds, let ξ and ξ1 be C k vector fields on M and
let η, η1 be C k vector fields on N . If ξ is ϕ-related to
ξ1 and η is ϕ-related to η1, then [ξ, η] is ϕ-related to
[ξ1, η1].
6.4. SUBMANIFOLDS, IMMERSIONS, EMBEDDINGS 371
6.4 Submanifolds, Immersions, Embeddings
Although the notion of submanifold is intuitively rather

clear, technically, it is a bit tricky.
In fact, the reader may have noticed that many different

definitions appear in books and that it is not obvious at
first glance that these definitions are equivalent.
What is important is that a submanifold, N of a given

manifold, M , not only have the topology induced M but
also that the charts of N be somewhow induced by those
of M .
(Recall that if X is a topological space and Y is a subset of

X, then the subspace topology on Y or topology induced
by X on Y has for its open sets all subsets of the form
Y ∩ U , where U is an arbitary subset of X.).
Given m, n, with 0 ≤ m ≤ n, we can view Rm as a

subspace of Rn using the inclusion
Rm ∼= Rm × {(0, . . . , 0)} ,→ Rm × Rn−m = Rn,
| {z }
n−m
given by
(x1, . . . , xm) 7→ (x1, . . . , xm, 0,
| .{z
. . , 0}).
n−m
Definition 6.4.1 Given a C k -manifold, M , of dimen-

sion n, a subset, N , of M is an m-dimensional subman-
ifold of M (where 0 ≤ m ≤ n) iff for every point, p ∈ N ,
there is a chart, (U, ϕ), of M , with p ∈ U , so that
ϕ(U ∩ N ) = ϕ(U ) ∩ (Rm × {0n−m}).
(We write 0n−m = (0, . . . , 0).)
| {z }
n−m
The subset, U ∩N , of Definition 6.4.1 is sometimes called

a slice of (U, ϕ) and we say that (U, ϕ) is adapted to N
(See O’Neill [?] or Warner [?]).
Other authors, including Warner [?], use the term sub-

manifold in a broader sense than us and they use the
word embedded submanifold for what is defined in Defi-
nition 6.4.1.
The following proposition has an almost trivial proof but

it justifies the use of the word submanifold:
Proposition 6.4.2 Given a C k -manifold, M , of di-

mension n, for any submanifold, N , of M of dimen-
sion m ≤ n, the family of pairs (U ∩ N, ϕ U ∩ N ),
where (U, ϕ) ranges over the charts over any atlas for
M , is an atlas for N , where N is given the subspace
topology. Therefore, N inherits the structure of a C k -
manifold.
In fact, every chart on N arises from a chart on M in the

following precise sense:
Proposition 6.4.3 Given a C k -manifold, M , of di-

mension n and a submanifold, N , of M of dimension
m ≤ n, for any p ∈ N and any chart, (W, η), of N at
p, there is some chart, (U, ϕ), of M at p so that
ϕ(U ∩ N ) = ϕ(U ) ∩ (Rm × {0n−m})
and
ϕ U ∩ N = η U ∩ N,
where p ∈ U ∩ N ⊆ W .
It is also useful to define more general kinds of “subman-

ifolds”.
Definition 6.4.4 Let ϕ: N → M be a C k -map of man-

ifolds.
(a) The map ϕ is an immersion of N into M iff dϕp is
injective for all p ∈ N .
(b) The set ϕ(N ) is an immersed submanifold of M iff
ϕ is an injective immersion.
(c) The map ϕ is an embedding of N into M iff ϕ is

an injective immersion such that the induced map,
N −→ ϕ(N ), is a homeomorphism, where ϕ(N )
is given the subspace topology (equivalently, ϕ is an
open map from N into ϕ(N ) with the subspace topol-
ogy). We say that ϕ(N ) (with the subspace topology)
is an embedded submanifold of M .
(d) The map ϕ is a submersion of N into M iff dϕp is
surjective for all p ∈ N .
Again, we warn our readers that certain authors (such

as Warner [?]) call ϕ(N ), in (b), a submanifold of M !
We prefer the terminology immersed submanifold .
The notion of immersed submanifold arises naturally in

the framewok of Lie groups.
Indeed, the fundamental correspondence between Lie groups

and Lie algebras involves Lie subgroups that are not nec-
essarily closed.
But, as we will see later, subgroups of Lie groups that are

also submanifolds are always closed.
It is thus necessary to have a more inclusive notion of

submanifold for Lie groups and the concept of immersed
submanifold is just what’s needed.
Immersions of R into R3 are parametric curves and im-

mersions of R2 into R3 are parametric surfaces. These
have been extensively studied, for example, see DoCarmo
[?], Berger and Gostiaux [?] or Gallier [?].
Immersions (i.e., subsets of the form ϕ(N ), where N is

an immersion) are generally neither injective immersions
(i.e., subsets of the form ϕ(N ), where N is an injective
immersion) nor embeddings (or submanifolds).
For example, immersions can have self-intersections, as

the plane curve (nodal cubic): x = t2 − 1; y = t(t2 − 1).
Injective immersions are generally not embeddings (or

submanifolds) because ϕ(N ) may not be homeomorphic
to N .
An example is given by the Lemniscate of Bernoulli, an

injective immersion of R into R2:
t(1 + t2)
x = 4
,
1+t
t(1 − t2)
y = .
1 + t4
There is, however, a close relationship between submani-

folds and embeddings.
Proposition 6.4.5 If N is a submanifold of M , then

the inclusion map, j: N → M , is an embedding. Con-
versely, if ϕ: N → M is an embedding, then ϕ(N )
with the subspace topology is a submanifold of M and
ϕ is a diffeomorphism between N and ϕ(N ).
In summary, embedded submanifolds and (our) subman-

ifolds coincide.
Some authors refer to spaces of the form ϕ(N ), where ϕ

is an injective immersion, as immersed submanifolds.
However, in general, an immersed submanifold is not a

submanifold.
One case where this holds is when N is compact, since

then, a bijective continuous map is a homeomorphism.
Our next goal is to review and promote to manifolds some

standard results about ordinary differential equations.
6.5. INTEGRAL CURVES, FLOW, ONE-PARAMETER GROUPS 379
6.5 Integral Curves, Flow of a Vector Field,

One-Parameter Groups of Diffeomorphisms
We begin with integral curves and (local) flows of vector

fields on a manifold.
Definition 6.5.1 Let ξ be a C k−1 vector field on a C k -

manifold, M , (k ≥ 2) and let p0 be a point on M . An
integral curve (or trajectory) for ξ with initial condi-
tion p0 is a C p−1-curve, γ: I → M , so that
γ̇(t) = ξ(γ(t)), for all t ∈ I and γ(0) = p0,
where I = ]a, b[ ⊆ R is an open interval containing 0.
What definition 6.5.1 says is that an integral curve, γ,

with initial condition p0 is a curve on the manifold M
passing through p0 and such that, for every point p = γ(t)
on this curve, the tangent vector to this curve at p, i.e.,
γ̇(t), coincides with the value, ξ(p), of the vector field ξ
at p.
Given a vector field, ξ, as above, and a point p0 ∈ M ,

is there an integral curve through p0? Is such a curve
unique? If so, how large is the open interval I?
We provide some answers to the above questions below.

manifold, M , (k ≥ 2) and let p0 be a point on M . A
local flow for ξ at p0 is a map,
ϕ: J × U → M,
where J ⊆ R is an open interval containing 0 and U is an
open subset of M containing p0, so that for every p ∈ U ,
the curve t 7→ ϕ(t, p) is an integral curve of ξ with initial
condition p.
Thus, a local flow for ξ is a family of integral curves for all

points in some small open set around p0 such that these
curves all have the same domain, J, independently of the
initial condition, p ∈ U .
The following theorem is the main existence theorem of

local flows.
This is a promoted version of a similar theorem in the

classical theory of ODE’s in the case where M is an open
subset of Rn.
Theorem 6.5.3 (Existence of a local flow) Let ξ be a

C k−1 vector field on a C k -manifold, M , (k ≥ 2) and
let p0 be a point on M . There is an open interval,
J ⊆ R, containing 0 and an open subset, U ⊆ M ,
containing p0, so that there is a unique local flow,
ϕ: J × U → M , for ξ at p0. Furthermore, ϕ is C k−1.
Theorem 6.5.3 holds under more general hypotheses, namely,

when the vector field satisfies some Lipschitz condition,
see Lang [?] or Berger and Gostiaux [?].
Now, we know that for any initial condition, p0, there is

some integral curve through p0.
However, there could be two (or more) integral curves

γ1: I1 → M and γ2: I2 → M with initial condition p0.
This leads to the natural question: How do γ1 and γ2

differ on I1 ∩ I2? The next proposition shows they don’t!
Proposition 6.5.4 Let ξ be a C k−1 vector field on a

C k -manifold, M , (k ≥ 2) and let p0 be a point on M .
If γ1: I1 → M and γ2: I2 → M are any two integral
curves both with initial condition p0, then γ1 = γ2 on
I1 ∩ I2.
Proposition 6.5.4 implies the important fact that there is

a unique maximal integral curve with initial condition
p.
Indeed, if {γk : Ik → M }k∈K is the family of all integral

curves with initial condition
S p (for some big index set,
K), if we let I(p) = k∈K Ik , we can define a curve,
γp: I(p) → M , so that
γp(t) = γk (t), if t ∈ Ik .
Since γk and γl agree on Ik ∩ Il for all k, l ∈ K, the curve

γp is indeed well defined and it is clearly an integral curve
with initial condition p with the largest possible domain
(the open interval, I(p)).
The curve γp is called the maximal integral curve with

initial condition p and it is also denoted γ(t, p).
Note that Proposition 6.5.4 implies that any two distinct

integral curves are disjoint, i.e., do not intersect each
other.
The following interesting question now arises: Given any

p0 ∈ M , if γp0 : I(p0) → M is the maximal integral curve
with initial condition p0, for any t1 ∈ I(p0), and if p1 =
γp0 (t1) ∈ M , then there is a maximal integral curve,
γp1 : I(p1) → M , with initial condition p1.
What is the relationship between γp0 and γp1 , if any? The

answer is given by
Proposition 6.5.5 Let ξ be a C k−1 vector field on

a C k -manifold, M , (k ≥ 2) and let p0 be a point
on M . If γp0 : I(p0) → M is the maximal integral
curve with initial condition p0, for any t1 ∈ I(p0), if
p1 = γp0 (t1) ∈ M and γp1 : I(p1) → M is the maximal
integral curve with initial condition p1, then
I(p1) = I(p0)−t1 and γp1 (t) = γγp0 (t1)(t) = γp0 (t+t1),
for all t ∈ I(p0) − t1
It is useful to restate Proposition 6.5.5 by changing point

of view.
So far, we have been focusing on integral curves, i.e., given

any p0 ∈ M , we let t vary in I(p0) and get an integral
curve, γp0 , with domain I(p0).
Instead of holding p0 ∈ M fixed, we can hold t ∈ R fixed

and consider the set
Dt(ξ) = {p ∈ M | t ∈ I(p)},
i.e., the set of points such that it is possible to “travel for
t units of time from p” along the maximal integral curve,
γp, with initial condition p (It is possible that Dt(ξ) = ∅).
By definition, if Dt(ξ) 6= ∅, the point γp(t) is well defined,

and so, we obtain a map,
Φξt : Dt(ξ) → M , with domain Dt(ξ), given by
Φξt (p) = γp(t).
The above suggests the following definition:


manifold, M , (k ≥ 2). For any t ∈ R, let
Dt(ξ) = {p ∈ M | t ∈ I(p)}
and
D(ξ) = {(t, p) ∈ R × M | t ∈ I(p)}
and let Φξ : D(ξ) → M be the map given by
Φξ (t, p) = γp(t).
The map Φξ is called the (global) flow of ξ and D(ξ) is
called its domain of definition. For any t ∈ R such that
Dt(ξ) 6= ∅, the map, p ∈ Dt(ξ) 7→ Φξ (t, p) = γp(t), is
denoted by Φξt (i.e., Φξt (p) = Φξ (t, p) = γp(t)).
Observe that
[
D(ξ) = (I(p) × {p}).
p∈M
Also, using the Φξt notation, the property of Proposition

6.5.5 reads
Φξs ◦ Φξt = Φξs+t, (∗)
whenever both sides of the equation make sense.
Indeed, the above says

Φξs (Φξt (p)) = Φξs (γp(t)) = γγp(t)(s) = γp(s + t) = Φξs+t(p).
Using the above property, we can easily show that the Φξt
are invertible. In fact, the inverse of Φξt is Φξ−t.
We summarize in the following proposition some addi-

tional properties of the domains D(ξ), Dt(ξ) and the
maps Φξt :
Theorem 6.5.7 Let ξ be a C k−1 vector field on a C k -

manifold, M , (k ≥ 2). The following properties hold:
(a) For every t ∈ R, if Dt(ξ) 6= ∅, then Dt(ξ) is open
(this is trivially true if Dt(ξ) = ∅).
(b) The domain, D(ξ), of the flow, Φξ , is open and the
flow is a C k−1 map, Φξ : D(ξ) → M .
(c) Each Φξt : Dt(ξ) → D−t(ξ) is a C k−1-diffeomorphism
with inverse Φξ−t.
(d) For all s, t ∈ R, the domain of definition of
Φξs ◦Φξt is contained but generally not equal to Ds+t(ξ).
However, dom(Φξs ◦ Φξt ) = Ds+t(ξ) if s and t have
the same sign. Moreover, on dom(Φξs ◦Φξt ), we have
Φξs ◦ Φξt = Φξs+t.
The reason for using the terminology flow in referring to

the map Φξ can be clarified as follows:
For any t such that Dt(ξ) 6= ∅, every integral curve, γp,

with initial condition p ∈ Dt(ξ), is defined on some open
interval containing [0, t], and we can picture these curves
as “flow lines” along which the points p flow (travel) for
a time interval t.
Then, Φξ (t, p) is the point reached by “flowing” for the

amount of time t on the integral curve γp (through p)
starting from p.
Intuitively, we can imagine the flow of a fluid through

M , and the vector field ξ is the field of velocities of the
flowing particles.
Given a vector field, ξ, as above, it may happen that

Dt(ξ) = M , for all t ∈ R.
In this case, namely, when D(ξ) = R × M , we say that

the vector field ξ is complete.
Then, the Φξt are diffeomorphisms of M and they form a

group.
The family {Φξt }t∈R a called a 1-parameter group of ξ.
In this case, Φξ induces a group homomorphism,

(R, +) −→ Diff(M ), from the additive group R to the
group of C k−1-diffeomorphisms of M .
By abuse of language, even when it is not the case that

Dt(ξ) = M for all t, the family {Φξt }t∈R is called a local
1-parameter group of ξ, even though it is not a group,
because the composition Φξs ◦ Φξt may not be defined.
When M is compact, it turns out that every vector field

is complete, a nice and useful fact.
Proposition 6.5.8 Let ξ be a C k−1 vector field on a

C k -manifold, M , (k ≥ 2). If M is compact, then ξ
is complete, i.e., D(ξ) = R × M . Moreover, the map
t 7→ Φξt is a homomorphism from the additive group R
to the group, Diff(M ), of (C k−1) diffeomorphisms of
M.
Remark: The proof of Proposition 6.5.8 also applies

when ξ is a vector field with compact support (this means
that the closure of the set {p ∈ M | ξ(p) 6= 0} is com-
pact).
A point p ∈ M where a vector field vanishes, i.e.,

ξ(p) = 0, is called a critical point of ξ.
Critical points play a major role in the study of vec-

tor fields, in differential topology (e.g., the celebrated
Poincaré–Hopf index theorem) and especially in Morse
theory, but we won’t go into this here (curious readers
should consult Milnor [?], Guillemin and Pollack [?] or
DoCarmo [?], which contains an informal but very clear
presentation of the Poincaré–Hopf index theorem).
Another famous theorem about vector fields says that

every smooth vector field on a sphere of even dimension
(S 2n) must vanish in at least one point (the so-called
“hairy-ball theorem”.
On S 2, it says that you can’t comb your hair without

having a singularity somewhere. Try it, it’s true!).
Let us just observe that if an integral curve, γ, passes

through a critical point, p, then γ is reduced to the point
p, i.e., γ(t) = p, for all t.
Then, we see that if a maximal integral curve is defined

on the whole of R, either it is injective (it has no self-
intersection), or it is simply periodic (i.e., there is some
T > 0 so that γ(t + T ) = γ(t), for all t ∈ R and γ is
injective on [0, T [ ), or it is reduced to a single point.
We conclude this section with the definition of the Lie

derivative of a vector field with respect to another vector
field.
Say we have two vector fields ξ and η on M . For any

p ∈ M , we can flow along the integral curve of ξ with
initial condition p to Φξt (p) (for t small enough) and then
evaluate η there, getting η(Φξt (p)).
Now, this vector belongs to the tangent space TΦξ (p)(M ),

t
but η(p) ∈ Tp(M ).
So to “compare” η(Φξt (p)) and η(p), we bring back η(Φξt (p))

to Tp(M ) by applying the tangent map, dΦξ−t, at Φξt (p),
to η(Φξt (p)) (Note that to alleviate the notation, we use
the slight abuse of notation dΦξ−t instead of d(Φξ−t)Φξ (p).)
t
Then, we can form the difference dΦξ−t(η(Φξt (p))) − η(p),

divide by t and consider the limit as t goes to 0.
This is the Lie derivative of η with respect to ξ at p,

denoted (Lξ η)p, and given by
dΦξ−t(η(Φξt (p))) − η(p)
(Lξ η)p = lim =
t−→0 t
d ξ ξ

(dΦ−t(η(Φt (p))) .
dt t=0
It can be shown that (Lξ η)p is our old friend, the Lie
bracket, i.e.,
(Lξ η)p = [ξ, η]p.

Tangent Spaces

Uploaded by

Copyright:

Available Formats

Tangent Spaces

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tangent Spaces

Uploaded by

Copyright:

Available Formats

Chapter 6

Manifolds, Tangent Spaces, Cotangent

In a previous Chapter we defined the notion of a manifold

In order to maximize the range of applications of the the-

The basic idea is still that, whatever a manifold is, it is

Of course, manifolds would be very dull without functions

This is a general fact learned from experience: Geometry

In particular, we still would like to “do calculus” on our

The small drawback with the more general approach is

We can still define the notion of a curve on a manifold,

Instead, we have to resort to the notion of chart. This is

For example, a geography atlas gives a set of maps of var-

Given Rn, recall that the projection functions,

For technical reasons, from now on, all topological spaces

Definition 6.1.1 Given a topological space, M , a chart

For any p ∈ M , a chart, (U, ϕ), is a chart at p iff p ∈ U .

The inverse, (Ω, ϕ−1), of a chart is called a

Given any two charts, (U1, ϕ1) and (U2, ϕ2), if

Clearly, ϕij = (ϕji )−1.

This is good news! Indeed, the whole arsenal of calculus

Definition 6.1.2 Given a topological space, M , and

(3) Whenever Ui ∩ Uj 6= ∅, the transition map ϕji (and

We must insure that we have enough charts in order to

For this, we must be able to add new charts whenever

Technically, given a C k n-atlas, A, on M , for any other

Two atlases A and A0 on M are compatible iff every

This is equivalent to saying that the union of the two

It is immediately verified that compatibility induces an

In fact, given an atlas, A, for M , the collection, A,

Definition 6.1.3 Given any two integers, n ≥ 1 and

Remark: It might have been better to use the terminol-

We can allow k = 0 in the above definitions. Condition

In this case, M is called a topological manifold of di-

We do not require a manifold to be connected but we

Actually, on every connected component of M , it can be

What happens if n = 0? In this case, every one-point

Observe that since Rn is locally compact and locally con-

Definition 6.1.4 Given a set, M , and any two integers,

(3) Whenever Ui ∩ Uj 6= ∅, the set ϕi(Ui ∩ Uj ) is open

But, this time, we give M the topology in which the

It is not difficult to verify that the axioms of a topology

It can also be shown that when M is equipped with the

One can also define the topology on M in terms of any

We also require M to be Hausdorff and second-countable

If the underlying topological space of a manifold is com-

Also, if A is some atlas for M and (U, ϕ) is a chart in A,

Thus, (V, ϕ  V ) is also a chart for M . This observation

Example 1. The sphere S n.

Thus, if we let UN = S n − {N } and US = S n − {S},

Furthermore, it is easily checked that on the overlap,

Example 2. The projective space RPn.

To define an atlas on RPn it is convenient to view RPn

It is customary to write (x1: · · · : xn+1) instead of

For any i, with 1 ≤ i ≤ n + 1, let

We can define a homeomorphism, ϕi, of Ui onto Rn, as

We can also define the maps, ψi, from Rn to Ui ⊆ RPn,

One easily checks that ϕi and ψi are mutual inverses, so

Intuitively, the space RPn is obtained by glueing the open

Example 3. The Grassmannian G(k, n).

Thus, (V, ϕ V ) is also a chart for M . This observation